How to write a good
Journal of Solid State Circuits
paper
Bram Nauta
JSSC Editor in Chief
University of Twente, The Netherlands
b.nauta@utwente.nl
http://icd.ewi.utwente.nl
November 2008
1
[A-SSCC 2008, Fukuoka, Japan]
Outline
• About the Journal
– Yesterday and today
– Journal versus Conference
– Organization of the Journal
– What kind of papers?
• Writing tips
– Paper outline
– General tips
• How NOT to write a JSSC paper
2
Outline
• About the Journal
– Yesterday and today
– Journal versus Conference
– Organization of the Journal
– What kind of papers?
• Writing tips
– Paper outline
– General tips
• How NOT to write a JSSC paper
3
History
• Established in 1966
– 4 issues per year
• Volume 1, nr 1: ISSCC 1966 issue
• Purpose= archive
4
TODAY
• The Nr 1 IEEE Journal
• By far: most downloaded IEEE Journal
• By far: most cited in all US Patents
– Over all technical disciplines
5
References in patents from top 25
companies to top 20 publishers
6
Top cited IEEE Journals in patents
Rank
1
2
3
4
5
6
7
8
9
Title
IEEE Journal of Solid-State Circuits
IEEE Transactions on Electron Devices
IEEE Transactions on Communications
IEEE Photonics Technology Letters
Journal of Lightwave Technology
Proceedings of the IEEE
IEEE Transactions on Magnetics
IEEE Transactions on Computing
IEEE J on Selected Areas in Communications
Cites
14,765
8,824
8,678
8,383
5,989
5,338
5,071
4,393
4,148
7
Outline
• About the Journal
– Yesterday and today
– Journal versus Conference
– Organization of the Journal
– What kind of papers?
• Writing tips
– Paper outline
– General tips
• How NOT to write a JSSC paper
8
Conferences
• Fast publication
• Usually a smaller idea
– Benchmark: Known circuit in new technology
– Smaller trick can be o.k.
– depends on conference
• Just accept or reject; no rewrite
– It may be incomplete
– It may lack key references
• Good for networking and Q&A
• SSCS conferences are available on IEEExplore
9
Journal publication
• Academic reputation
– Journals can have more status than conferences
• Depends on the field
• Reviewed Journal gives a “quality stamp”
– Reviewers demand corrections & clarifications
• Archive your work
– Wider scope
– More theory
– More technical information
– More Educational
– More references
10
Outline
• About the Journal
– Yesterday and today
– Journal versus Conference
– Organization of the Journal
– What kind of papers?
• Writing tips
– Paper outline
– General tips
• How NOT to write a JSSC paper
11
Organization regular JSSC papers
Editor
Associate Editor
Reviewers
Author
12
Associate Editors
•
•
•
•
•
•
•
•
•
•
Baas, Bevan
Behzad, Arya
Flynn, Michael
Gharpurey, Ranjit
Gillingham, Peter
Halonen, Kari
Karanicolas, Andrew
Kim, Beomsup,
Leenaerts, Domine
Liu, Shen-Iuan,
Mok, Philip
Nairn David
Natarajan Shreedhar
Rusu, Stefan
Razavi, Behzad,
Savoj, Jafar
Shaeffer, Derek
Shepard, ken
Young, Darrin
13
Special Issues on conferences
•
•
•
•
•
•
•
•
•
December
January
April
May
July
August
September
October
November
ISSCC-analog, RF
ISSCC-dig+rest
VLSI
RFIC
ESSCIRC
CICC
BCTM
CSIC
A-SSC
(issue)
(issue)
(issue)
(section)
(issue)
(issue)
(section)
(section)
(section)
14
Organization special Issue
Editor
Associate Editor
Reviewers
Author
Guest Editor
Reviewers
15
Procedure (regular papers)
•
•
•
•
Author submits manuscript to Editor
Editor sends to Associate editor
Associate Editor sends to reviewers
Associate Editor makes decision:
– AWR accept with revisions ☺
– REJ reject
– REF refer to other Journal
Its not a democratic process!!!!!!!!!!!
16
Procedure (regular papers)
• Author rewrites
– Give list of how you changed manuscript based on
reviewers comments
– Do not reply to reviewer but change your manuscript
• Associate Editor makes final decision
• OR: Associate Editor organizes second review
• Author submits final package to Associate Editor
• Associate Editor inspects and forwards to Editor
• Editor compiles issues and sends to IEEE
17
Time schedule (regular papers)
• Delay to first decision
• Delay author rewrite
• Publication delay
~100 days
~100 days
~140 days
• Total delay
~340 days
18
Origin of regular papers
• 365 submissions/year: 1/day
– Asia
– USA
– EU
– ROW
40% (Taiwanese Universities)
25%
25%
10%
• Same regional distribution as ISSCC
• 90% is analog/RF!!
19
Accept/reject regular papers
•
•
•
•
Accept
Reject
Refer to other Journal
Withdrawn
36%
58%
3%
3%
• Main reject reasons:
– Not enough novelty/innovation
– Not enough news w.r.t. prepublication
20
Outline
• About the Journal
– Yesterday and today
– Journal versus Conference
– Organization of the Journal
– What kind of papers?
• Writing tips
– Paper outline
– General tips
• How NOT to write a JSSC paper
21
What papers are good for JSSC?
• IC Implementation oriented
– Not: microwave PCP/ modeling/ device only
• Not too much theory ( Better use TCAS)
• Integrated Circuit needed
• But to have an IC is not enough!
The work must be of HIGH Quality
22
What is high quality?
• Must describe an INNOVATIVE IDEA
– Not just a permutation
– Not a know circuit in new technology
– Not a combination of known techniques
– But a real new concept / invention
• AND idea must be proven on IC
• AND idea must significantly advance
state-of- art
23
Prepublication
• ~90% of submitted papers have
prepublication at conference(s)
– That’s ok
• Conferences are available on IEEExplore
• So a JSSCC paper should add significant
information to the prepublications
24
Prepublication policy
• “A JSSCC paper should be worth
reading for a person who has read the
conference paper.”
– Better description of state-of-art
– More theory + proof of concept
– More measurements
– More discussion (mismatch, no-idealities etc)
– More benchmark, more discussion
– More references
25
Prepublication policy
• NOT each conference paper deserves to
be a JSSC paper
• NOR each work without prepublication
deserves to be a JSSC paper
• It’s a delicate thing, but all about QUALITY
26
Outline
• About the Journal
– Yesterday and today
– Journal versus Conference
– Organization of the Journal
– What kind of papers?
• Writing tips
– Paper outline
– General tips
• How NOT to write a JSSC paper
27
A Typical paper outline
•
•
•
•
•
•
•
•
Title
Abstract
Introduction
Body
Experimental results
Discussion
Conclusion
references
28
The Title
• Must describe the paper
• Not too vague
– “A novel receiver”
• Do not use “novel” anyway
– “5-GHz RF Frontends for Ultra-Low-Voltage and UltraLow-Power Operations”
• How much is is Ultra?
• But exactly what is really new:
– “Noise canceling technique for wideband receivers”
• Or exactly what is achieved
– “A 1.5GHz 1.3dB NF, 10mW down converter in 65nm
CMOS for GPS applications”
• Or both!
29
Abstract
•
•
•
•
1 paragraph
Exactly what paper is about
Can have overlap with conclusions
Keywords, indexing terms
– Use many!!
– So your paper can be found
– You will be cited a lot
30
Introduction
• Describe the problem you solve
–
–
–
–
Open the subject
Zoom in step by step
Describe your assumptions
Each step is one paragraph
• Describe the state-of-art
– Use plenty of references
• Tell basic your idea
– This motivates the reader to continue
– Cite your prepublications and tell the difference
• Give outline
31
The body
• Explain your key idea
• Build up Step by step
– One thinking step at the time
– Each step is one paragraph
• Proof that it makes sense
– Use mathematics
– Give exactly your boundary conditions
– Give results in comprehensive way
32
The body
• Be self-critical and realistic:
– does it really make sense?
• E.g. for a linearity improvement technique:
– If power dissipation is larger
– And noise is also larger
– And you know that P~SNR: does this make sense?
• Is it just the technology or your smartness?
– E.g. speed ~fT or fmax
• Are practical boundary conditions met?
– VCO @ high frequency but Pout=-30dBm
33
Experimental Results
• Describe exactly what has been measured and
how.
–
–
–
–
–
–
–
Describe setup
“Bio Biased”? (manual tweaking and tuning)
Probe or PCB?
What equipment?
How many samples?
PVT?
Batch to batch spread?
• Experiment must be repeatable and of practical
use (e.g. for industry)
34
Experimental Results
• Compare with theory / simulations
• Does it prove your idea and theory?
• Always tell if a result is measures. Simulated or
calculated.
“Figure x shows the noisefigure versus frequency”
is this measured? Simulated? Calculated?
Estimated?
35
IC realization
• Give chip photograph
– Dimensions
– What is what
• Give technology + options
36
Discuss results
• Compare to state-of-art in fair way
– Show all relevant data + papers
– A table can help
• although measurements are hard to compare
• Use common FoM definitions
– ADC, VCO, filter
• Be careful to define your own FoM
– Do not misuse FoM for showing off
– Power ~ SNR. BW makes sense
– Power/bondpad is NOT a good FoM!!
37
Discuss results
• Help the reader to interpret the results
• Absolute accuracy needed?
– show many samples, proof batch to batch
robustness
• Matching needed?
– show many samples
• Calibrated circuits?
– describe what input signal is used/required.
When does it go wrong? How realistic is it?
38
Discuss Results:
A useless PLL benchmark:
39
Conclusions
• Start writing with this
• First make a bullet list for yourself
– A hand full of bullets
– So you know where to write towards
– This gives your paper focus
• Conclusion should be readable without reading
the whole paper
• Convince the reader
What did we learn?
40
References
• Include latest state of the art
– For benchmark
• But also refer to the original papers
– Go back in time!
– Most references are younger than 5 years
– While most ideas are much older!
• Textbooks are useful too
41
Outline
• About the Journal
– Yesterday and today
– Journal versus Conference
– Organization of the Journal
– What kind of papers?
• Writing tips
– Paper outline
– General tips
• How NOT to write a JSSC paper
42
Writing tips:
A well written paper
gives the impression of
a good idea
43
Writing Tips
• If a paper is too complex:
– Reviewers don't understand it
– Reviewers don’t believe it
– Reviewers will not like it
• If a paper is too simplistic
– Reviewers think its nothing special
– Even if the results are good
44
General writing tips
• Make your problem relevant
• Start with the “big picture”
• Take the reader by the hand
– Step by step explanation
•
•
•
•
Highlight innovation
Do not give too many equations
Do not give too much theoretical details
Do not try to make a tutorial
45
General writing tips
• Do not use “very” but give the numbers
• Avoid to use “novel”
– everything you don´t cite should be novel
• Use short sentences
• Use simple words
• One point per paragraph
– First or last sentence is most important
– The rest is explanation
46
General writing tips
• If you are stuck:
– Tell a friend what you did.
– Use the words & slides like on your conference paper
– Polish the text later
• Let a fellow student read & comment
• Ask native speaker to correct language
• Polish, Polish, Polish
– Reviewers hate mistakes!!
– It iz raely anojing to raed tekst width misstakes
47
Figures
• Make the figures like a cartoon
– Reader can understand idea by looking at
figures + caption only
• Spend a lot of time to make good figures
– Papers with bad figures almost always get
rejected
48
Figures
• Must be readable in single column:
• Not good:
49
Outline
• About the Journal
– Yesterday and today
– Journal versus Conference
– Organization of the Journal
– What kind of papers?
• Writing tips
– Paper outline
– General tips
• How NOT to write a JSSC paper
50
Do NOT
• Publish the same material elsewhere
– Reviewers+readers always see this; its unethical
• Change your paper after acceptance and before
publication
– E.g. remove reference to competitor
– Reviewers always see this
• Use someone else´s ideas
– “Someone else” is reading too
• Hide “unpleasant” measurements
51
Do NOT
• Fabricate or falsify results
– Not tune bias for each measuring point
– Not make few chips and measure different
parameters on different chips
– Or even completely falsify results
52
Outline
• About the Journal
– Yesterday and today
– Journal versus Conference
– Organization of the Journal
– What kind of papers?
• Writing tips
– Paper outline
– General tips
• How NOT to write a JSSC paper
53
Summary
• IEEE Journal of Solid-State Circuit
– Most downloaded, most cited in patents
• Needs an Innovative new idea
– Working silicon is not enough
– Must improve state-of-art
• Needs new material after prepublication
• Reviewers are demanding
• Your writing technique can help
54